Please don't hard-code catch paths
yurivict opened this issue · 1 comments
Description
Instead of:
#include "../Catch2/single_include/catch2/catch.hpp"
it should be
#include <catch2/catch.hpp>
Then makefile should take care of the include path.
If the git submodule is present - it should be used. If not - `pkg-config --cflags catch2`
should be used.
And it doesn't make sense to run git from makefiles because in many cases this directory comes from a tarball and it isn't a git repository.
Then makefile should take care of the include path.
This makes sense to me.
And it doesn't make sense to run git from makefiles because in many cases this directory comes from a tarball and it isn't a git repository.
In this case - because we include catch2 as a git submodule with a specific commit - I think it makes sense to keep it here. Worth remembering that although this package has some C++ and may some day get more direct C++ use, for now 99% of its users interface with it as a Python package. Keeping the instructions for running the C++ tests simple and in a MakeFile is useful. At some point in the future we could consider a more sophisticated CMake infrastructure.